Responsible use

Remove AI Metadata is a privacy and inspection tool. It shows what an image file carries and removes stored metadata on your own device. This page explains what the tool is for, where the line is, and what removing metadata does not change. These rules are part of our Terms of Service.

What the cleaner is for

• Removing location and device data before you share a photo. Phone photos often carry GPS coordinates, the phone or camera model and the exact time the photo was taken. Removing them before you post a picture of your home, your children or your workplace is the most common use of the tool.
• Removing prompts and workflows from PNG files. Stable Diffusion, ComfyUI and similar tools store the full prompt, seed, model and node workflow inside the PNG. Removing them keeps your working method private. The image and any visible disclosure stay as they are.
• Fixing a real photo that was labeled as AI. A photo retouched with one generative tool, such as removing a cable with Generative Fill, can carry an AI record that makes a platform label the whole picture as AI. The scan shows you which record triggered the label, so you can decide what is accurate under that platform's rules. If the rules ask you to disclose the edit, disclose it.
• Protecting sources. Journalists, researchers and activists remove metadata from images they receive or publish, so the file does not reveal who took it, with which device or where.
• Smaller files. Embedded thumbnails, editing history, large XMP packets and C2PA manifests can add tens of kilobytes or more to an image.
• Security research and testing. Checking how a website, app or pipeline handles provenance metadata, testing your own C2PA implementation, or preparing test files.
• Checking before you trust. Seeing whether an image has Content Credentials, who signed them and whether they validate, before you rely on it or pass it on.

What it must not be used for

• Removing AI disclosures that the law requires. Under the EU AI Act, people who publish deepfakes (AI-generated or manipulated images, audio or video of real people, places or events that would falsely appear authentic) must disclose that the content is artificially generated or manipulated, with lighter rules for evidently artistic or satirical work (Article 50(4), applies from August 2, 2026). In China, the Measures for Labeling AI-Generated and Synthetic Content (in force since September 1, 2025) forbid deleting, altering, forging or hiding the required AI labels, and forbid providing tools for others to do so. If a law that applies to you requires a label, keep it.
• Removing labels that platform rules require. Several platforms require you to disclose realistic AI content: Meta for photorealistic video and realistic audio, TikTok for AI content that shows realistic scenes, YouTube for realistic altered or synthetic videos. Removing metadata does not change those duties.
• Breaking your AI provider's terms. Under the EU's voluntary Code of Practice on transparency of AI-generated content (final text June 10, 2026), providers such as OpenAI, Google, Meta, Microsoft and Anthropic committed to forbid intentional removal of their AI markings in their terms. Check the terms of the tool that made your image.
• Deepfake impersonation and deception. Do not use the tool to make AI-generated or manipulated images of real people look authentic, or for fraud, harassment, non-consensual intimate imagery, election disinformation or fake evidence.
• Hiding someone else's authorship. Do not remove copyright or author information from someone else's work to pass it off as yours or to infringe it.

What removing metadata does not change

• Invisible watermarks stay. Google SynthID is in the pixels of images from Google's Gemini app and, since May 19, 2026, of images OpenAI generates in ChatGPT, Codex and most of its API. TikTok began rolling out its own invisible watermark in November 2025. Removing metadata leaves all of these in place.
• Visible watermarks stay. Logos, signatures and text drawn on the image are part of the picture.
• A label may still appear. Platforms can also read invisible watermarks, rely on what the uploader discloses or run their own detection.
• Your obligations stay. Laws and platform rules apply to what you publish, whatever the file's metadata says.

How we put this into practice

• We describe the tool as a privacy and inspection tool. We never promise that it makes AI images undetectable or that it bypasses a platform.
• The scan shows Content Credentials before you remove them (the signer, the app that created them and whether they validate), so you can make an informed decision.
• Our Terms of Service say: if you publish realistic AI-generated content, you remain responsible for any disclosure required by law or platform rules; do not use the service to remove labels you are required to keep.
• Your images never reach our servers, so we cannot see how you use the tool. That is why we spell the rules out here.

To report misuse or ask a question, email support@removeaimetadata.com.

Last reviewed: October 20, 2026. This page is general information, not legal advice.